Hydrogeologic Processes Impacting Storage, Fate, and Transport of

Apr 14, 2016 - Department of Forest and Natural Resources Management, College of Environmental Science and Forestry, State University of New. York, On...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/est

Hydrogeologic Processes Impacting Storage, Fate, and Transport of Chloride from Road Salt in Urban Riparian Aquifers Sarah H. Ledford,*,†,§ Laura K. Lautz,† and John C. Stella‡ †

Department of Earth Sciences, Syracuse University, 204 Heroy Geology Laboratory, Syracuse, New York 13244, United States Department of Forest and Natural Resources Management, College of Environmental Science and Forestry, State University of New York, One Forestry Drive, Syracuse, New York 13210, United States



S Supporting Information *

ABSTRACT: Detrimental effects of road salt runoff on urban streams are compounded by its facilitated routing via storm drains, ditches, and flood channels. Elevated in-stream salinity may also result from seasonal storage and discharge of chloride in groundwater, and previous work has hypothesized that groundwater discharge to streams may have the effect of diluting stream chloride concentrations in winter and enriching them in summer. However, the hydrogeological processes controlling these patterns have not been thoroughly investigated. Our research focuses on an urban stream and floodplain system in Syracuse, NY, to understand how groundwater and surface water exchange impacts chloride storage, fate, and transport. We created a 3D groundwater flow and solute transport model of the floodplain, calibrated to the distributions of floodplain hydraulic heads and groundwater fluxes to the stream throughout the reach. We used a sensitivity analysis to calibrate and evaluate the influence of model parameters, and compared model outputs to field observations. The main source mechanism of chloride to the floodplain aquifer was highconcentration, overbank flood events in winter that directly recharged groundwater. The modeled residence time and storage capacity of the aquifer indicate that restoration projects designed to promote floodplain reconnection and the frequency of overbank flooding in winter have the potential to temporarily store chloride in groundwater, buffer surface water concentrations, and reduce stream concentrations following periods of road salting.



nonwinter months,10−12 and this increase is due to chloride retention within watersheds.13 There is a need to understand chloride dynamics in urban systems given the negative effects on aquatic and terrestrial organisms at high concentrations.14,15 The U.S. EPA has established acute and chronic ambient water quality limits for chloride of 860 mg/L and 230 mg/L, respectively.16 Groundwater storage is a major component of chloride fate and transport in urban areas.11,17−19 Groundwater chloride concentrations are impacted by chloride storage in soils,20,21 delivery of chloride from the unsaturated zone,22−24 and interaction with surface water.25 Controls on surface water chloride concentrations in winter and spring are primarily impacted by flushing of salted impervious surface cover.18,25,26 However, in streams receiving groundwater discharge, mixing of high salinity surface waters with low salinity groundwater decreases surface water concentrations in winter. Long-term salinization of floodplain groundwater is occurring throughout

INTRODUCTION Urbanization has a clear impact on surface water quality. Alteration of stream corridors in urban areas, including cement bank armoring, channelization, floodplain development, and degradation of the riparian zone, collectively disconnect surface waters from riparian areas and groundwater.1 Disconnection from groundwater, along with other hydrologic impacts of urbanization, results in decreased ecological function and ecosystem services provided by urban streams.1−3 Stream restoration increasingly emphasizes restoration of those ecosystem functions and services, by improving riparian habitat and modifying channel hydromorphology to promote floodplain reconnection.4 However, the effects of restoring hydrologic connectivity of streams and floodplains (i.e., the exchange of water between a stream and its riparian zone and floodplain aquifer) on chloride transport in areas impacted by road salt use are not well studied.5 In 2005 alone, 18 million megagrams of road salt were applied to U.S. roads for driver safety,6 along with unquantified amounts applied to private parking lots and sidewalks. Even low levels of impervious surface cover result in increases of chloride concentrations in surface waters.7−9 Baseflow chloride concentrations are increasing in urban streams, including during © XXXX American Chemical Society

Received: January 25, 2016 Revised: April 13, 2016 Accepted: April 14, 2016

A

DOI: 10.1021/acs.est.6b00402 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

impacted by road runoff and described in further detail in Ledford and Lautz.25 The lower 1.5 km of the stream meanders through a large cemetery before flowing into a riparian floodplain for the final 500 m. The floodplain, with its mature and extensive riparian vegetation, complex channel morphology, and hydrologic connection to the riparian aquifer, is the model site of this study. The area has a temperate climate, with an average annual snowfall of 315 cm and 98 cm of rain. Average monthly temperatures range from −4.7 °C in January to 21.8 °C in July.28 Since 2009, the Onondaga County Department of Transportation has used 100% salt on the roads for deicing.29 Field Methods. Five piezometers were installed in the floodplain in September 2011 in a transect perpendicular to the stream (Figure 2b). P1 was closest to the stream, at a distance of 1.1 m, with approximately equal spacing of piezometers through P5, which was 12 m from the stream (Table S1). Piezometers were installed by hand augering a borehole up to 1.5 m below the land surface. Piezometers were cased with polyvinyl chloride (PVC) pipe (1.9 cm diameter) and had a 30.5 cm sand-packed screen. Sand was used to fill voids between the piezometer and the borehole sides. Hydraulic head at each piezometer during stream baseflow was measured on October 5, 2011. Groundwater samples were collected from each piezometer on 20−22 different dates between June 8, 2012 until June 4, 2013, depending on whether the piezometer recovered quickly enough from purging to sample. Stream water samples were collected from the middle of the stream from May 11, 2012 until June 4, 2013 at 24 stations positioned longitudinally along the stream. Water samples were collected in 60 mL HDPE bottles, were stored at 4 °C, and were filtered within 24 h using Whatman GF/F 0.7 μm nominal pore size filters. Samples were analyzed for anion chemistry using a Dionex ICS-2000 Ion Chromatograph with five in-house standards for calibration and three U.S. Geological Survey standards for calibration verification. Measurement error was estimated as three times the standard deviation of replicate standard measurements. Wet precipitation data were collected on the green roof of the Syracuse Center of Excellence, approximately 2 km from the studied watershed (http://syracusecoe.syr.edu) and analyzed by ion chromatography (Driscoll, unpublished data). A Solinst LTC Levelogger Junior pressure transducer and conductivity logger was calibrated and installed adjacent to the piezometer transect on January 19, 2013 (Figure 2b). It recorded water height and conductivity at 10 min intervals from that date until July 22, 2014. Logger conductivity measurements were converted to chloride concentrations by an empirical relationship between conductivity and chloride from grab samples taken at the same site (see SI). Streambed elevations and piezometer casing elevations were surveyed using a Nikon Nivo Total Station. Groundwater discharge to the stream over the reach was measured using either a Sontek acoustic Doppler velocimeter (ADV) or by doing Rhodamine Water Tracer (RWT) or bromide injections over the reach. Road length and distribution data were downloaded from the NYS Office of Cyber Security.30 All analyses of road densities and other watershed characteristics were completed using ESRI ArcGIS. Weather data were collected from the Community Collaborative Rain, Hail & Snow Network (CoCoRaHS) station NY-OG-2 from July 1, 2012 until July 1, 2014.31 This station is located in the research

northern climates, but storage may help in the short term to mitigate potentially harmful effects of highly saline events. In this study, we used field observations in conjunction with groundwater modeling to investigate the hydrogeological processes controlling chloride fate and transport in the saturated zone of an urban floodplain. As our study system, we used an urban stream in New York State that is subject to heavy applications of road salts in winter (Figure 1). The

Figure 1. (a) Location of Meadowbrook Creek watershed within Onondaga County and in relationship to the City of Syracuse. (b) Meadowbrook Creek starts from a stormwater retention basin before flowing east into an Erie Canal feeder channel. The model domain is the floodplain located in the last 500 m of the stream.

modeling approach simulated floodplain hydrologic processes that could retain excess chloride from deicers and slowly discharge it year-round. These processes focused on spatial and temporal interactions between the five principle controls on chloride transport in urban riparian floodplains (Figure 2a): surface water-groundwater interactions; hillslope groundwater discharge; push and pull of water to and from floodplains with changes in stream stage (e.g., bank storage); precipitation recharge; and groundwater recharge during overbank flooding events. The only process that removed water and chloride from groundwater was discharge to the stream. Finally, we discuss the implications of our findings for stream restoration projects, specifically those that restore riparian zones and promote hydrologic connection between urban streams and adjacent groundwater systems.



MATERIALS AND METHODS Study Site. Meadowbrook Creek is a first-order urban stream in Syracuse and DeWitt, New York (Figure 1). The study reach flows east for 5.6 km until it discharges into the Erie Canal system.27 The upper 4.1 km of the stream is heavily B

DOI: 10.1021/acs.est.6b00402 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 2. (a) Conceptual model of how chloride could enter the floodplain. (b) The model is made of 7 layers, with two hydraulic conductivities, a constant head boundary along the river, a general head boundary along the hillslope break, and recharge. Stream height and specific conductivity were recorded at the water level logger. Model results were compared to chloride grab samples collected from the piezometer transect and at the two grab sample locations indicated.

The model was 10 m in depth and contains seven layers. Changes in land surface elevation longitudinally along the floodplain were defined based on the average slope of the river bottom from field surveys. The slope of the floodplain toward the river from the hillslope was based on the observed slope in land surface along a piezometer transect (slope =0.01). Layer thicknesses range from 0.20 to 2.5 m (Figure 2b), with thinner layers within the top 1.5 where we had more detailed field data. All model layers have a constant thickness throughout the model domain. Observations of drill cuttings during borehole augering show the floodplain sediments to be comprised of a surficial layer of silty-clay, underlain by a more permeable sandy-silt layer. The top three layers of the model domain were assigned a hydraulic conductivity two orders of magnitude lower than the bottom four layers of the model domain to represent the observed floodplain sediment structure. Conductivities of all layers were designated as anisotropic, with Kx,y:Kz of 10. Winkley27 found that outwash sand and gravel surficial deposits in the area have bulk conductivities ranging from 10−2 to 10−5 m/s and are highly permeable. Literature values for hydraulic conductivity of floodplain sediments were used as initial conductivity values for layers four through seven, which were then adjusted during steady-state model calibration (Table 1). The northern and southern boundaries of the model are located at the break in hillslope observed in the field and were

watershed, approximately 2.5 km from the model site. These data included precipitation, snowfall, and snow depth. Groundwater Flow Modeling. We aimed to simulate annual patterns in groundwater hydrology and solute transport that are the dominant factors regulating the fate and transport of road salt runoff in a typical year. The model excluded the unsaturated zone due to its minimal thickness (generally